
import pandas as pd
import numpy as np

import matplotlib.pyplot as plt

from sklearn.model_selection import train_test_split
from sklearn.model_selection import learning_curve

from sklearn.preprocessing import MinMaxScaler
from sklearn.preprocessing import label_binarize

import scipy.cluster.hierarchy as shc

from sklearn.feature_selection import SelectKBest
from sklearn.feature_selection import chi2

from sklearn.decomposition import PCA
import seaborn as sns

from sklearn.neural_network import MLPClassifier
from sklearn.svm import SVC
from sklearn.neighbors import KNeighborsClassifier

from sklearn.cluster import KMeans
from sklearn.cluster import DBSCAN
from sklearn_extra.cluster import KMedoids

from sklearn.metrics import classification_report
from sklearn.metrics import accuracy_score
from sklearn.metrics import silhouette_score
from sklearn.metrics import roc_curve
from sklearn.metrics import auc
from sklearn.metrics import log_loss

from mpl_toolkits.mplot3d import Axes3D
import scipy.cluster.hierarchy as shc

import warnings
warnings.filterwarnings('ignore')


data = pd.read_csv('ML_Project_data.csv')
modelData = data.drop(['label'], axis=1)
labels = data['label']


data.sample(n=5)


scaler = MinMaxScaler(feature_range=(0, 1)).fit(modelData)
modelData_scaled = scaler.transform(modelData)


pd.concat([pd.DataFrame(modelData_scaled), labels], axis=1).sample(n=5)


data.corr().sample(n=5)


display(data.corr()['label'])

f0      -0.096589
f1      -0.095564
f2       0.135957
f3      -0.118736
f4       0.146644
           ...   
f65      0.005227
f66      0.054075
f67      0.082596
f68      0.096932
label    1.000000
Name: label, Length: 70, dtype: float64


plt.figure(figsize = (20, 10))

data_log = data.copy()
data_log = np.log10(data_log)
data_heatmap = sns.heatmap(data_log.corr(), linewidths = 0.5)

plt.show()


print("Selected Features with correlation between features: \n")
data_log_corr = data_log.corr()['label'].sort_values(ascending = False)[1:51]
print(np.array(data_log_corr.index))

Selected Features with correlation between features: 

['f2' 'f4' 'f3' 'f1' 'f20' 'f50' 'f8' 'f23' 'f25' 'f0' 'f29' 'f11' 'f18'
 'f51' 'f66' 'f13' 'f49' 'f27' 'f10' 'f67' 'f9' 'f15' 'f6' 'f48' 'f68'
 'f21' 'f22' 'f53' 'f12' 'f38' 'f7' 'f65' 'f28' 'f40' 'f44' 'f36' 'f47'
 'f55' 'f16' 'f64' 'f14' 'f61' 'f33' 'f46' 'f31' 'f24' 'f59' 'f57' 'f32'
 'f54']


features = list(modelData.columns)
selectModel = SelectKBest(chi2, k=50)
selectModel.fit(modelData_scaled, labels)
selected_features = selectModel.get_feature_names_out(features)


print("Selected Features with feature selsection algorithm: \n")
display(selected_features)

Selected Features with feature selsection algorithm:

array(['f0', 'f1', 'f2', 'f3', 'f4', 'f5', 'f6', 'f7', 'f8', 'f9', 'f10',
       'f11', 'f12', 'f13', 'f14', 'f15', 'f16', 'f17', 'f18', 'f19',
       'f20', 'f23', 'f24', 'f25', 'f26', 'f27', 'f28', 'f29', 'f30',
       'f31', 'f33', 'f34', 'f37', 'f40', 'f41', 'f42', 'f44', 'f46',
       'f50', 'f52', 'f56', 'f60', 'f61', 'f62', 'f63', 'f64', 'f65',
       'f66', 'f67', 'f68'], dtype=object)


print("Features that are in feature selsection algorithm and not in correlation selsection: \n")
print(list(set(selected_features) - set(data_log_corr.index)))

Features that are in feature selsection algorithm and not in correlation selsection: 

['f56', 'f17', 'f37', 'f30', 'f19', 'f41', 'f62', 'f42', 'f5', 'f34', 'f63', 'f52', 'f60', 'f26']


deleted_cols = dict()
selected_cols = []

for i in range(len(selected_features)):
    col = int(selected_features[i][1:])
    deleted_cols[col] = 0

for i in range(len(modelData.columns)):
    if i not in deleted_cols:
        selected_cols.append(i)

data_model_scaled = np.delete(modelData_scaled, selected_cols, 1)


trainData, testData, trainLabels, testLabels = train_test_split(data_model_scaled, labels, test_size=0.2,  random_state=42)


sns.displot(trainLabels, kde=False)

<seaborn.axisgrid.FacetGrid at 0x1ea166d38e0>


sns.displot(testLabels, kde=False)

<seaborn.axisgrid.FacetGrid at 0x1ea15968c70>


print("Train Data Size: ", len(trainData))
print("Test Data Size: ", len(testData))

Train Data Size:  9792
Test Data Size:  2448


pca_2 = PCA(n_components = 2)

trainData_principal_2 = pd.DataFrame(pca_2.fit_transform(trainData, trainLabels), columns = ['P1', 'P2'])
trainLabels_principal_2 = pd.DataFrame(trainLabels).sort_index(ignore_index=True)
trainData_principal_2 = pd.concat([trainData_principal_2, trainLabels_principal_2], axis=1)


testData_principal_2 = pd.DataFrame(pca_2.fit_transform(testData, testLabels), columns = ['P1', 'P2'])
testLabels_principal_2 = pd.DataFrame(testLabels).sort_index(ignore_index=True)
testData_principal_2 = pd.concat([testData_principal_2, testLabels_principal_2], axis=1)


dataClass0 = trainData_principal_2.loc[trainData_principal_2['label']==0]
dataClass1 = trainData_principal_2.loc[trainData_principal_2['label']==1]
dataClass2 = trainData_principal_2.loc[trainData_principal_2['label']==2]
dataClass3 = trainData_principal_2.loc[trainData_principal_2['label']==3]
dataClass4 = trainData_principal_2.loc[trainData_principal_2['label']==4]

plt.figure(figsize = (6, 6)) 

plt.scatter(dataClass0['P1'], dataClass0['P2'], c='red', marker='.')
plt.scatter(dataClass1['P1'], dataClass1['P2'], c='blue', marker='.')
plt.scatter(dataClass2['P1'], dataClass2['P2'], c='green', marker='.')
plt.scatter(dataClass3['P1'], dataClass3['P2'], c='yellow', marker='.')
plt.scatter(dataClass4['P1'], dataClass4['P2'], c='purple', marker='.')

plt.show()


print("Dendrogram for TrainData: \n")
plt.figure(figsize =(8, 8))
trainDendrogram = shc.dendrogram((shc.linkage(trainData_principal_2, method ='ward')))

Dendrogram for TrainData:


print("Dendrogram for TestData: \n")
plt.figure(figsize =(8, 8))
testDendrogram = shc.dendrogram((shc.linkage(testData_principal_2, method ='ward')))

Dendrogram for TestData:


def print_report(model):
    print("TrainData: ")
    print(classification_report(trainLabels, model.predict(trainData)))
    print()
    print("TestData: ")
    print(classification_report(testLabels, model.predict(testData)))


def visualize_roc_curve(model, data_in, labels_in, n_classes):
    Y_true = label_binarize(labels_in, classes=[int(i) for i in range(n_classes)])
    Y_score = label_binarize(model.predict(data_in), classes=[int(i) for i in range(n_classes)])
    
    fpr = dict()
    tpr = dict()
    roc_auc = dict()

    for i in range(n_classes):
        fpr[i], tpr[i], _ =  roc_curve(Y_true[:, i], Y_score[:, i])
        roc_auc[i] =  auc(fpr[i], tpr[i])

    plt.figure()

    colors = ['red', 'blue', 'green', 'yellow', 'purple']

    for i in range(n_classes):    
        temp_label = 'ROC of class {0} (area={1:0.2f})'.format(i, roc_auc[i])
        plt.plot(fpr[i], tpr[i], label=temp_label, color=colors[i])

        plt.plot([0, 1], [0, 1], linestyle='--', color='black')
        plt.xlim([-0.05, 1.05])
        plt.ylim([-0.05, 1.05])
        plt.xlabel('False Positive Rate')
        plt.ylabel('True Positive Rate')
        plt.title('ROC curve')
        plt.legend(loc="lower right")
    plt.show()


KNN = KNeighborsClassifier(n_neighbors=2, weights='distance', metric='manhattan', algorithm='ball_tree')
KNN.fit(trainData, trainLabels)

KNeighborsClassifier(algorithm='ball_tree', metric='manhattan', n_neighbors=2,
                     weights='distance')


print("KNN Classifier Train accuracy: ", KNN.score(trainData, trainLabels)*100)
print("KNN Classifier Test accuracy: ", KNN.score(testData, testLabels)*100)

KNN Classifier Train accuracy:  100.0
KNN Classifier Test accuracy:  90.890522875817


print("KNN classification report: \n")
print_report(KNN)

KNN classification report: 

TrainData: 
              precision    recall  f1-score   support

           0       1.00      1.00      1.00      1985
           1       1.00      1.00      1.00      1849
           2       1.00      1.00      1.00      2077
           3       1.00      1.00      1.00      2104
           4       1.00      1.00      1.00      1777

    accuracy                           1.00      9792
   macro avg       1.00      1.00      1.00      9792
weighted avg       1.00      1.00      1.00      9792


TestData: 
              precision    recall  f1-score   support

           0       0.92      0.93      0.92       512
           1       0.91      0.91      0.91       472
           2       0.90      0.91      0.91       529
           3       0.93      0.89      0.91       521
           4       0.87      0.91      0.89       414

    accuracy                           0.91      2448
   macro avg       0.91      0.91      0.91      2448
weighted avg       0.91      0.91      0.91      2448


print("KNN Classifier: ROC curve of train")
visualize_roc_curve(KNN, trainData, trainLabels, n_classes=5)

KNN Classifier: ROC curve of train


print("KNN Classifier: ROC curve of test")
visualize_roc_curve(KNN, testData, testLabels, n_classes=5)

KNN Classifier: ROC curve of test


train_knn_accuracy = list()
test_knn_accuracy = list()

for i in range(1, 20):
    temp_KNN = KNeighborsClassifier(n_neighbors=i, weights='distance', metric='manhattan', algorithm='ball_tree')
    temp_KNN.fit(trainData, trainLabels)
    train_knn_accuracy.append(temp_KNN.score(trainData, trainLabels))
    test_knn_accuracy.append(temp_KNN.score(testData, testLabels))
    
plt.figure(figsize = (5, 5)) 
plt.plot([int(i) for i in range(1, 20)], train_knn_accuracy, color="y", label="KNN train data accuracy")
plt.plot([int(i) for i in range(1, 20)], test_knn_accuracy, color="b", label="KNN test data accuracy")

plt.title("Find best number of k for KNN classifier", color = 'green', fontweight = 'bold') 
plt.xlabel("K numbers", color = 'green', fontweight = 'bold') 
plt.ylabel("Accuracy score", color = 'green', fontweight = 'bold')

plt.legend(loc="best")
plt.show()


knn_train_sizes, knn_train_accuracy, knn_testt_accuracy = learning_curve(KNN, trainData, trainLabels, scoring="accuracy")

plt.figure(figsize = (5, 5)) 
plt.plot(knn_train_sizes, knn_train_accuracy.mean(1), "o-", color="y", label="KNN train data accuracy")
plt.plot(knn_train_sizes, knn_testt_accuracy.mean(1), "o-", color="b", label="KNN test data accuracy")

plt.title("Learning Curve of KNN classifier", color = 'green', fontweight = 'bold') 
plt.xlabel("Training examples", color = 'green', fontweight = 'bold') 
plt.ylabel("Accuracy score", color = 'green', fontweight = 'bold')

plt.legend(loc="best")
plt.show()


knn_train_sizes, knn_train_loss, knn_test_loss = learning_curve(KNN, trainData, trainLabels, scoring="neg_log_loss")

plt.figure(figsize = (5, 5)) 
plt.plot(knn_train_sizes, -knn_train_loss.mean(1), "o-", color="y", label="KNN train data loss")
plt.plot(knn_train_sizes, -knn_test_loss.mean(1), "o-", color="b", label="KNN test data loss")

plt.title("Learning Curve of KNN classifier", color = 'green', fontweight = 'bold') 
plt.xlabel("Training examples", color = 'green', fontweight = 'bold') 
plt.ylabel("Loss score", color = 'green', fontweight = 'bold')

plt.legend(loc="best")
plt.show()


svm = SVC(kernel='rbf', C=2, gamma=8, decision_function_shape='ovo', random_state=42)
svm.fit(trainData, trainLabels)

SVC(C=2, decision_function_shape='ovo', gamma=8, random_state=42)


print("SVM Classifier Train accuracy: ", svm.score(trainData, trainLabels)*100)
print("SVM Classifier Test accuracy: ", svm.score(testData, testLabels)*100)

SVM Classifier Train accuracy:  99.98978758169935
SVM Classifier Test accuracy:  87.13235294117648


print("SVM classification report: \n")
print_report(svm)

SVM classification report: 

TrainData: 
              precision    recall  f1-score   support

           0       1.00      1.00      1.00      1985
           1       1.00      1.00      1.00      1849
           2       1.00      1.00      1.00      2077
           3       1.00      1.00      1.00      2104
           4       1.00      1.00      1.00      1777

    accuracy                           1.00      9792
   macro avg       1.00      1.00      1.00      9792
weighted avg       1.00      1.00      1.00      9792


TestData: 
              precision    recall  f1-score   support

           0       0.94      0.87      0.90       512
           1       0.94      0.83      0.88       472
           2       0.90      0.88      0.89       529
           3       0.73      0.91      0.81       521
           4       0.91      0.85      0.88       414

    accuracy                           0.87      2448
   macro avg       0.88      0.87      0.87      2448
weighted avg       0.88      0.87      0.87      2448


print("SVM Classifier: ROC curve of train")
visualize_roc_curve(svm, trainData, trainLabels, n_classes=5)

SVM Classifier: ROC curve of train


print("SVM Classifier: ROC curve of test")
visualize_roc_curve(svm, testData, testLabels, n_classes=5)

SVM Classifier: ROC curve of test


train_svm_accuracy = list()
test_svm_accuracy = list()

for i in range(1, 10):
    temp_SVM = SVC(kernel='rbf', C=2, gamma=i, decision_function_shape='ovo', random_state=42)
    temp_SVM.fit(trainData, trainLabels)
    train_svm_accuracy.append(temp_SVM.score(trainData, trainLabels))
    test_svm_accuracy.append(temp_SVM.score(testData, testLabels))
    
plt.figure(figsize = (5, 5)) 
plt.plot([int(i) for i in range(1, 10)], train_svm_accuracy, color="y", label="KNN train data accuracy")
plt.plot([int(i) for i in range(1, 10)], test_svm_accuracy, color="b", label="KNN test data accuracy")

plt.title("Find best value of gamma for SVM classifier", color = 'green', fontweight = 'bold') 
plt.xlabel("gamma value", color = 'green', fontweight = 'bold') 
plt.ylabel("Accuracy score", color = 'green', fontweight = 'bold')

plt.legend(loc="best")
plt.show()


train_svm_accuracy = list()
test_svm_accuracy = list()

for i in range(1, 10):
    temp_SVM = SVC(kernel='rbf', C=i, gamma=8, decision_function_shape='ovo', random_state=42)
    temp_SVM.fit(trainData, trainLabels)
    train_svm_accuracy.append(temp_SVM.score(trainData, trainLabels))
    test_svm_accuracy.append(temp_SVM.score(testData, testLabels))
    
plt.figure(figsize = (5, 5)) 
plt.plot([int(i) for i in range(1, 10)], train_svm_accuracy, color="y", label="KNN train data accuracy")
plt.plot([int(i) for i in range(1, 10)], test_svm_accuracy, color="b", label="KNN test data accuracy")

plt.title("Find best value of C for SVM classifier", color = 'green', fontweight = 'bold') 
plt.xlabel("C value", color = 'green', fontweight = 'bold') 
plt.ylabel("Accuracy score", color = 'green', fontweight = 'bold')

plt.legend(loc="best")
plt.show()


svm_train_sizes, svm_train_accuracy, svm_testt_accuracy = learning_curve(svm, trainData, trainLabels, scoring="accuracy")

plt.figure(figsize = (5, 5)) 
plt.plot(svm_train_sizes, svm_train_accuracy.mean(1), "o-", color="y", label="SVM train data accuracy")
plt.plot(svm_train_sizes, svm_testt_accuracy.mean(1), "o-", color="b", label="SVM test data accuracy")

plt.title("Learning Curve of SVM classifier", color = 'green', fontweight = 'bold') 
plt.xlabel("Training examples", color = 'green', fontweight = 'bold') 
plt.ylabel("Accuracy score", color = 'green', fontweight = 'bold')


plt.legend(loc="best")
plt.show()


svm_train_sizes, svm_train_loss, svm_test_loss = learning_curve(KNN, trainData, trainLabels, scoring="neg_log_loss")

plt.figure(figsize = (5, 5)) 
plt.plot(svm_train_sizes, -svm_train_loss.mean(1), "o-", color="y", label="SVM train data loss")
plt.plot(svm_train_sizes, -svm_test_loss.mean(1), "o-", color="b", label="SVM test data loss")

plt.title("Learning Curve of SVM classifier", color = 'green', fontweight = 'bold') 
plt.xlabel("Training examples", color = 'green', fontweight = 'bold') 
plt.ylabel("loss score", color = 'green', fontweight = 'bold')

plt.legend(loc="best")
plt.show()


MLP = MLPClassifier(hidden_layer_sizes = (128, 64, 32, 8), batch_size = 32, solver = 'sgd', random_state=4, \
                      verbose=False, momentum=0.85, max_iter=350, learning_rate_init = 0.01, n_iter_no_change=100)
MLP.fit(trainData, trainLabels)

MLPClassifier(batch_size=32, hidden_layer_sizes=(128, 64, 32, 8),
              learning_rate_init=0.01, max_iter=350, momentum=0.85,
              n_iter_no_change=100, random_state=4, solver='sgd')


print("MLP Classifier Train accuracy: ", MLP.score(trainData, trainLabels)*100)
print("MLP Classifier Test accuracy: ", MLP.score(testData, testLabels)*100)

MLP Classifier Train accuracy:  99.97957516339869
MLP Classifier Test accuracy:  80.6781045751634


print("MLP classification report: \n")
print_report(MLP)

MLP classification report: 

TrainData: 
              precision    recall  f1-score   support

           0       1.00      1.00      1.00      1985
           1       1.00      1.00      1.00      1849
           2       1.00      1.00      1.00      2077
           3       1.00      1.00      1.00      2104
           4       1.00      1.00      1.00      1777

    accuracy                           1.00      9792
   macro avg       1.00      1.00      1.00      9792
weighted avg       1.00      1.00      1.00      9792


TestData: 
              precision    recall  f1-score   support

           0       0.84      0.85      0.85       512
           1       0.85      0.79      0.82       472
           2       0.78      0.81      0.80       529
           3       0.80      0.78      0.79       521
           4       0.76      0.80      0.78       414

    accuracy                           0.81      2448
   macro avg       0.81      0.81      0.81      2448
weighted avg       0.81      0.81      0.81      2448


print("MLP Classifier: ROC curve of train")
visualize_roc_curve(MLP, trainData, trainLabels, n_classes=5)

MLP Classifier: ROC curve of train


print("MLP Classifier: ROC curve of test")
visualize_roc_curve(MLP, testData, testLabels, n_classes=5)

MLP Classifier: ROC curve of test


mlp_train_sizes, mlp_train_accuracy, mlp_test_accuracy = learning_curve(MLP, trainData, trainLabels, scoring="accuracy")

plt.figure(figsize = (5, 5)) 

plt.plot(mlp_train_sizes, mlp_train_accuracy.mean(1), "o-", color="y", label="MLP train data accuracy")
plt.plot(mlp_train_sizes, mlp_test_accuracy.mean(1), "o-", color="b", label="MLP test data accuracy")

plt.title("Learning Curve of MLP classifier", color = 'green', fontweight = 'bold') 
plt.xlabel("Training examples", color = 'green', fontweight = 'bold') 
plt.ylabel("Accuracy score", color = 'green', fontweight = 'bold') 

plt.legend(loc="best")
plt.show()


mlp_train_sizes, mlp_train_loss, mlp_test_loss = learning_curve(MLP, trainData, trainLabels, scoring="neg_log_loss")

plt.figure(figsize = (5, 5)) 
plt.plot(mlp_train_sizes, -mlp_train_loss.mean(1), "o-", color="y", label="MLP train data loss")
plt.plot(mlp_train_sizes, -mlp_test_loss.mean(1), "o-", color="b", label="MLP test data loss")

plt.title("Learning Curve of MLP classifier", color = 'green', fontweight = 'bold') 
plt.xlabel("Training examples", color = 'green', fontweight = 'bold') 
plt.ylabel("Loss score", color = 'green', fontweight = 'bold') 

plt.legend(loc="best")
plt.show()


plt.figure(figsize = (6, 6)) 

plt.plot(MLP.loss_curve_, color="r", label="KNN train data loss")

plt.title("Learning Curve of MLP classifier", color = 'green', fontweight = 'bold') 
plt.xlabel("Epoch", color = 'green', fontweight = 'bold') 
plt.ylabel("Loss score", color = 'green', fontweight = 'bold') 

plt.legend(loc="best")
plt.show()


def visualize_clustring_result(model, feature1, feature2, data_in, text1, text2):
    print(text1, "\n")
    fig = plt.figure(figsize=(6, 6))
    plt.scatter(feature1, feature2, c=model.fit_predict(data_in), cmap ='rainbow')
    plt.show()
    
    print()
    print(text2, "\n")
    fig = plt.figure(figsize=(6, 6))
    Axes3D(fig).scatter(feature1, feature2, zs=0, c=model.fit_predict(data_in),  cmap ='rainbow')
    plt.show()


def run_KMeans_cluster(k=5):
    print("For clusster number: ", k)
    print()
    kMeans = KMeans(n_clusters=k, init='k-means++', random_state=2, algorithm='elkan', max_iter=100)
    kMeans.fit(trainData_principal_2, trainLabels)
    print("kMeans train accuracy score: ", accuracy_score(np.array(trainLabels), kMeans.predict(trainData_principal_2))*100)
    print("kMeans train silhouette score: ", silhouette_score(trainData_principal_2, kMeans.predict(trainData_principal_2), random_state=2)*100)
    print("kMeans test accuracy score: ", accuracy_score(np.array(testLabels), kMeans.predict(testData_principal_2))*100)
    print("kMeans test silhouette score: ", silhouette_score(testData_principal_2, kMeans.predict(testData_principal_2), random_state=2)*100)
    print()
    text1 = "kMeans Cluster: result of train in 2D"
    text2 = "kMeans Cluster: result of train in 3D"
    visualize_clustring_result(kMeans, trainData_principal_2['P1'], trainData_principal_2['P2'], trainData, text1, text2)
    text1 = "kMeans Cluster: result of test in 2D"
    text2 = "kMeans Cluster: result of test in 3D"
    visualize_clustring_result(kMeans, testData_principal_2['P1'], testData_principal_2['P2'], testData, text1, text2)


for i in range(2, 6):
    run_KMeans_cluster(i)

For clusster number:  2

kMeans train accuracy score:  19.904003267973856
kMeans train silhouette score:  56.08980633583279
kMeans test accuracy score:  20.465686274509803
kMeans test silhouette score:  56.245800451094865

kMeans Cluster: result of train in 2D

kMeans Cluster: result of train in 3D

kMeans Cluster: result of test in 2D

kMeans Cluster: result of test in 3D

For clusster number:  3

kMeans train accuracy score:  20.567810457516337
kMeans train silhouette score:  47.56765320629847
kMeans test accuracy score:  20.179738562091504
kMeans test silhouette score:  47.843915962551726

kMeans Cluster: result of train in 2D

kMeans Cluster: result of train in 3D

kMeans Cluster: result of test in 2D

kMeans Cluster: result of test in 3D

For clusster number:  4

kMeans train accuracy score:  19.985702614379086
kMeans train silhouette score:  49.38724823671695
kMeans test accuracy score:  20.547385620915033
kMeans test silhouette score:  49.80377961166413

kMeans Cluster: result of train in 2D

kMeans Cluster: result of train in 3D


dbscan = DBSCAN(eps=0.8, min_samples=3, metric='manhattan', n_jobs=2)
dbscan.fit(trainData_principal_2, trainLabels)

DBSCAN(eps=0.8, metric='manhattan', min_samples=3, n_jobs=2)


print("DBSCAN train accuracy score: ", accuracy_score(np.array(trainLabels), dbscan.fit_predict(trainData_principal_2))*100)
print("DBSCAN train silhouette score: ", silhouette_score(trainData_principal_2, dbscan.fit_predict(trainData_principal_2), random_state=2)*100)

DBSCAN train accuracy score:  20.26143790849673
DBSCAN train silhouette score:  53.37338778709126


print("DBSCAN test accuracy score: ", accuracy_score(np.array(testLabels), dbscan.fit_predict(testData_principal_2))*100)
print("DBSCAN test silhouette score: ", silhouette_score(testData_principal_2, dbscan.fit_predict(testData_principal_2), random_state=2)*100)

DBSCAN test accuracy score:  21.07843137254902
DBSCAN test silhouette score:  53.79870289337694


text1 = "DBSCAN Cluster: result of train in 2D"
text2 = "DBSCAN Cluster: result of train in 3D"

visualize_clustring_result(dbscan, trainData_principal_2['P1'], trainData_principal_2['P2'], trainData, text1, text2)

DBSCAN Cluster: result of train in 2D

DBSCAN Cluster: result of train in 3D


text1 = "DBSCAN Cluster: result of test in 2D"
text2 = "DBSCAN Cluster: result of test in 3D"

visualize_clustring_result(dbscan, testData_principal_2['P1'], testData_principal_2['P2'], testData, text1, text2)

DBSCAN Cluster: result of test in 2D

DBSCAN Cluster: result of test in 3D


def run_kMedoids_cluster(k=5):
    print("For clusster number: ", k)
    print()
    kMedoids = KMedoids(n_clusters=k, init='k-medoids++', random_state=2, method='pam', max_iter=100)
    kMedoids.fit(trainData_principal_2, trainLabels)
    print("kMedoids train accuracy score: ", accuracy_score(np.array(trainLabels), kMedoids.predict(trainData_principal_2))*100)
    print("kMedoids train silhouette score: ", silhouette_score(trainData_principal_2, kMedoids.predict(trainData_principal_2), random_state=2)*100)
    print("kMedoids test accuracy score: ", accuracy_score(np.array(testLabels), kMedoids.predict(testData_principal_2))*100)
    print("kMedoids test silhouette score: ", silhouette_score(testData_principal_2, kMedoids.predict(testData_principal_2), random_state=2)*100)
    print()
    text1 = "kMedoids Cluster: result of train in 2D"
    text2 = "kMedoids Cluster: result of train in 3D"
    visualize_clustring_result(kMedoids, trainData_principal_2['P1'], trainData_principal_2['P2'], trainData, text1, text2)
    text1 = "kMedoids Cluster: result of test in 2D"
    text2 = "kMedoids Cluster: result of test in 3D"
    visualize_clustring_result(kMedoids, testData_principal_2['P1'], testData_principal_2['P2'], testData, text1, text2)


for i in range(2, 6):
    run_kMedoids_cluster(i)

For clusster number:  2

kMedoids train accuracy score:  19.904003267973856
kMedoids train silhouette score:  56.08980633583279
kMedoids test accuracy score:  20.465686274509803
kMedoids test silhouette score:  56.245800451094865

kMedoids Cluster: result of train in 2D

kMedoids Cluster: result of train in 3D

kMedoids Cluster: result of test in 2D

kMedoids Cluster: result of test in 3D

For clusster number:  3

kMedoids train accuracy score:  20.567810457516337
kMedoids train silhouette score:  45.5947139975777
kMedoids test accuracy score:  20.79248366013072
kMedoids test silhouette score:  45.57884509722503

kMedoids Cluster: result of train in 2D

kMedoids Cluster: result of train in 3D

kMedoids Cluster: result of test in 2D

kMedoids Cluster: result of test in 3D

For clusster number:  4

kMedoids train accuracy score:  19.730392156862745
kMedoids train silhouette score:  48.96910171371724
kMedoids test accuracy score:  19.444444444444446
kMedoids test silhouette score:  49.37024168807112

kMedoids Cluster: result of train in 2D

kMedoids Cluster: result of train in 3D

	f0	f1	f2	f3	f4	f5	f6	f7	f8	f9	...	f60	f61	f62	f63	f64	f65	f66	f67	f68	label
4710	88191	2735.590516	205493.1480	5711.057133	1.127761e+06	0.332884	0.413978	0.306309	0.457731	0.318386	...	60.617756	46.782078	55.462086	49.677887	39.311531	38.796692	42.473606	38.756374	39.476429	1
10407	87860	2878.920663	290295.7266	6092.454826	1.198869e+06	0.220314	0.249701	0.400222	0.312943	0.531078	...	68.645561	88.746216	67.703003	61.458874	75.024498	69.949928	77.711037	80.714333	126.083107	4
6716	89611	2832.788390	849805.1446	5981.480497	2.491899e+06	0.397320	0.301977	0.498761	0.425237	0.207440	...	84.415504	59.736752	83.095253	69.194260	48.553337	53.077389	51.408463	47.567596	79.228539	2
326	129489	3197.852430	236779.6748	6408.194584	5.906822e+05	0.444583	0.404323	0.392631	0.371602	0.452789	...	80.655640	70.033463	91.528915	65.181450	72.857300	68.044167	76.253075	63.906807	134.757660	0
2314	100269	2757.664710	301362.2994	5959.980627	8.864881e+05	0.277739	0.321861	0.473990	0.495933	0.597967	...	70.380203	54.794155	48.231483	59.453911	47.974365	38.566723	49.718597	70.830276	52.456680	0

	0	1	2	3	4	5	6	7	8	9	...	60	61	62	63	64	65	66	67	68	label
1798	0.276517	0.491690	0.096484	0.669311	0.109886	0.259546	0.219901	0.265585	0.394037	0.318401	...	0.114109	0.109921	0.141366	0.094607	0.160881	0.075514	0.094319	0.083526	0.185056	0
8231	0.316442	0.537036	0.354929	0.673425	0.274790	0.703290	0.493119	0.431580	0.380109	0.396719	...	0.155184	0.089495	0.168392	0.127533	0.094582	0.077410	0.060683	0.068158	0.106952	3
5509	0.254443	0.368984	0.350453	0.449949	0.375661	0.191690	0.307691	0.275375	0.432937	0.264305	...	0.065793	0.040994	0.036944	0.056227	0.050705	0.075249	0.109416	0.061545	0.094089	2
1665	0.368799	0.563075	0.150143	0.672924	0.172755	0.603566	0.524089	0.693137	0.465763	0.388934	...	0.087799	0.060319	0.084824	0.050963	0.027194	0.033785	0.028158	0.055130	0.042921	0
3402	0.157449	0.233064	0.038211	0.269756	0.057347	0.383860	0.165143	0.142548	0.078045	0.138757	...	0.065538	0.027733	0.055470	0.057295	0.033495	0.040531	0.035610	0.037421	0.042005	1

	f0	f1	f2	f3	f4	f5	f6	f7	f8	f9	...	f60	f61	f62	f63	f64	f65	f66	f67	f68	label
f65	0.021948	-0.065246	-0.000571	-0.131971	0.017111	-0.221510	-0.238824	-0.215850	-0.188895	-0.131188	...	0.488392	0.554746	0.613382	0.674763	0.779231	1.000000	0.788892	0.657977	0.565096	0.005227
f58	-0.057363	-0.067362	0.175110	-0.080524	0.177137	-0.080030	-0.069983	-0.066445	-0.033953	-0.057506	...	0.582815	0.555936	0.509971	0.423429	0.361593	0.355301	0.310409	0.329624	0.342647	-0.042850
f20	-0.014651	-0.006215	0.066500	-0.004373	0.039209	-0.080859	0.102574	0.006602	0.652443	-0.101997	...	0.063532	0.040562	0.005658	0.004053	0.007841	-0.014684	0.039689	0.056916	0.075549	0.070595
f42	0.148930	0.271169	0.110282	0.293202	0.061979	0.325655	0.242169	0.144765	0.105987	0.081936	...	0.082181	0.091303	0.128421	0.158061	0.098076	0.007480	-0.052973	-0.145493	-0.207338	-0.074216
f32	0.016674	-0.113782	-0.145599	-0.129720	-0.191805	0.103361	0.062053	0.087142	0.043098	0.045576	...	-0.028008	-0.032266	-0.037649	-0.031447	-0.042148	-0.050465	-0.076629	-0.071055	-0.063937	-0.006073

Import libraries¶

Read Data¶

Data Normalization¶

Features Selection¶

Train Test Splitting¶

Reduce The Dimensionality With PCA¶

Classifying¶

KNN classifier¶

SVM classifier¶

MLP classifier¶

Clustering¶

K Means cluster¶

DBSCAN cluster¶

K medoids cluster¶